How do i clear model loaded using stable-ts from the ram memory? (both whisper and faster whisper)

[Dadangdut33]

I have tried using these but it does not seem to work, can anyone help? or is it not possible to clear it from memory?

# for whisper model
import stable_whisper
import torch
import gc
from time import sleep

x = stable_whisper.load_model("base")
print("loaded model")
sleep(5)

print("deleting from memory")
del x.encoder
del x.decoder

torch.cuda.empty_cache()
gc.collect()
print("cleared")

sleep(5)
input("done")

# for faster whisper
import stable_whisper
import torch
import gc
from time import sleep

x = stable_whisper.load_faster_whisper("base")
print("loaded model")
sleep(5)

print("deleting from memory")
del x.model

torch.cuda.empty_cache()
gc.collect()
print("cleared")

sleep(5)
input("done")

[Dadangdut33]

After trying it around a bit more in my code i found out that the memory might actually be reserved for cached for when we want to load another or the same model. Because when i first load small model and be done with it, it will just stay there taking up the space in the ram, but after that when i load medium model, the memory will go up until i finish doing task with it and then the memory will go back down to how it previously was after loading the small model.